Article Text

In early rheumatoid arthritis, anticitrullinated peptide antibodies associate with low number of affected joints and rheumatoid factor associates with systemic inflammation
  1. Eleftheria Pertsinidou1,2,
  2. Saedis Saevarsdottir3,4,
  3. Vivek Anand Manivel1,
  4. Lars Klareskog5,
  5. Lars Alfredsson6,7,
  6. Linda Mathsson-Alm1,2,
  7. Monika Hansson5,
  8. Martin Cornillet8,
  9. Guy Serre8,
  10. Rikard Holmdahl9,
  11. Karl Skriner10,
  12. Per-Johan Jakobsson5,
  13. Helga Westerlind3,
  14. Johan Askling3,
  15. Johan Rönnelid1
  1. 1 Department of Immunology Genetics and Pathology, Uppsala University, Uppsala, Sweden
  2. 2 ImmunoDiagnostics Division, Thermo Fisher Scientific, Uppsala, Sweden
  3. 3 Clinical Epidemiology Division, Department of Medicine, Karolinska Institutet, Solna, Stockholm, Sweden
  4. 4 Faculty of Medicine, School of Health Sciences, University of Iceland, Reykjavik, Iceland
  5. 5 Rheumatology Unit, Department of Medicine, Karolinska Institute, Solna, Stockholm, Sweden
  6. 6 Institute of Environmental Medicine, Karolinska Institute, Stockholm, Sweden
  7. 7 Center for Occupational and Environmental Medicine, Region Stockholm, Stockholm, Sweden
  8. 8 Institut Toulousain des Maladies Infectieuses et Inflammatoires, UMR1291 Inserm, 5051 CNRS, Université de Toulouse 3, Toulouse, France
  9. 9 Department of Medical Inflammation Research, Karolinska Institute, Stockholm, Sweden
  10. 10 Department of Medicine, Charité University Hospital, Berlin, Germany
  1. Correspondence to Mrs Eleftheria Pertsinidou, Department of Immunology Genetics and Pathology, Uppsala University, Uppsala, Sweden; eleftheria.pertsinidou{at}


Objectives To investigate how individual rheumatoid arthritis (RA) autoantibodies associate with individual signs and symptoms at the time of RA diagnosis.

Methods IgA, IgG, IgM rheumatoid factor (RF), antibodies against cyclic citrullinated peptide version 2 (anti-CCP2) and 16 individual antibodies against citrullinated protein (ACPA) reactivities were analysed centrally in baseline sera from 1600 patients with RA classified according to the 1987 American College of Rheumatology (ACR) criteria. These results were related to C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), number of swollen and tender joints (SJC and TJC), 28-joint disease activity scores (DAS28 and DAS28CRP), global disease activity evaluated by the patients and Health Assessment Questionnaire, all obtained at baseline.

Results Individually, all autoantibodies except immunoglobulin G (IgG) RF associated with low SJC and TJC and with high ESR. In IgM RF-negative patients, ACPA associated strictly with low number of swollen and tender joints. This association persisted in multiple regression and stratified analyses where IgM and IgA RF instead associated with inflammation expressed as ESR. Among subjects without any ACPA peptide reactivity, there was no association between RF isotypes and ESR. The effect of RF on ESR increased with the number of ACPA reactivities, especially for IgM RF. In patients fulfilling the 1987 ACR criteria without taking RF into account, associations between IgM RF and high ESR, as well as between ACPA and low joint counts, remained.

Conclusion Whereas ACPA associate with low counts of affected joints in early RA, RF associates with elevated measures of systemic inflammation in an ACPA-dependent manner. This latter finding corroborates in vitro models of ACPA and RF in immune complex-induced inflammation. These phenotypic associations are independent of classification criteria.

  • Arthritis, Rheumatoid
  • Autoantibodies
  • Rheumatoid Factor
  • Anti-Citrullinated Protein Antibodies
  • Autoimmune Diseases

Data availability statement

Data are available upon reasonable request.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Two previous studies have reported that occurrence of rheumatoid factor (RF) or antibodies against citrullinated proteins (ACPAs) not only conveys a more severe future prognosis but also associates with less disease activity and inflammation in early RA.

  • One study reported that RF but not ACPA associates with elevated disease activity in early RA.


  • Whereas ACPA associate with low number of swollen and tender joints in early RA, immunoglobulin M (IgM) and IgA RF associate with elevated ESR and CRP, but only in the presence of ACPA.

  • Thus, RF and ACPA associate with different individual 28-joint disease activity scores (DAS28) components in opposite directions.


  • The impact of different RA-associated autoantibodies should be assessed individually, and different DAS28 components should be evaluated separately as outcome markers.

  • This approach may increase the precision in algorithms for serology-based treatment decisions in early RA.


Autoantibody serology including rheumatoid factor (RF) and antibodies against citrullinated proteins (ACPAs) is an important component in rheumatoid arthritis (RA) diagnosis. Previous studies have shown that RF is associated with an aggressive RA phenotype.1 More recent studies have described that especially ACPAs are strongly associated with erosive disease.2 3

The association between RF, ACPA and the clinical phenotype at RA diagnosis may harbour important clues to the pathogenesis of the disease as well as the predictive capacity of RF and ACPA. In a study on the prognosis in early arthritis, Barra et al found higher 28-joint disease activity scores (DAS28) at RA diagnosis, based on the ACR/European League Against Rheumatism (EULAR) 2010 classification criteria, among patients negative for both RF and ACPA.4 Using the same RA classification criteria, Nordberg et al, reported higher levels of inflammation as defined by clinical measures and ultrasound in patients negative for both ACPA and RF. The authors hypothesised that the difference between seropositive and double seronegative patients might reflect the high number of affected joints required for seronegative patients to fulfil the 2010 ACR/EULAR classification criteria.5 When Aletaha et al investigated baseline data from four RA trials, they found that patients positive for RF had higher disease activity regardless of ACPA status, whereas ACPA-positive patients (regardless of RF status) had similar or lower disease activity than ACPA-negative patients.6 In all of the three studies, the autoantibody measurements had been performed locally at different sites (instead of subject to centralised testing) and merged for analyses, and none of the studies investigated the impact of individual immune globulin isotypes of RF.

Already the first paper defining ACPA described the substantial heterogeneity across citrullinated epitopes between individual patients with RA.7 Using a planar multiplex array investigating 16 ACPA specificities in 2825 patients with RA from the Epidemiological Investigations in RA (EIRA) case–control study, we demonstrated that anti-CCP2-negative RA patients who had ACPA as determined by our multiplex array showed the same associations to shared epitope and smoking as those previously described for anti-CCP2-positive RA.8 This assay is therefore a more stringent (as compared with anti-CCP2) way to identify purely ACPA-negative RA patients. In the previous studies, no clinical evaluation of the association of these ACPAs and aspects of RA disease activity were performed, and data on RF analyses were not included.9

In the present study, we therefore used the ACPA multiplex array data together with central analysis of immunoglobulin A (IgA), IgG and IgM RF as well as anti-CCP2 to assess the association of these antibodies, and combinations thereof, with different clinical RA phenotypes at the time of RA diagnosis.

Subjects and methods


The study population was the same 2825 patients with RA from the EIRA Study as in our above-mentioned study on autoantibodies, shared epitope (SE) and smoking, newly diagnosed with RA between 1996 and 2010.8 Inclusion criteria in the current study were a clinical RA diagnosis, fulfilment of the 1987 ACR classification criteria, 18–70 years of age, less than 12 months of arthritis symptoms and less than 40 days between RA diagnosis with clinical characterisation and blood sampling for autoantibody serology.10 Following exclusion of patients with missing linked clinical data from the Swedish Rheumatology Quality registry (SRQ; n=505) and patients with more than 1 year of symptom duration at EIRA inclusion or more than 40 days between the RA diagnosis and blood sampling (n=720), 1600 patients remained for analysis (figure 1A). Basic information on EIRA has been published previously.11 12 By the time of inclusion (work-up for RA), all patients had been classified as seropositive or seronegative according to RF measured in their local hospitals. At the time of inclusion, nephelometry or turbidometry, that is, non-isotype specific techniques which primarily detect IgM RF, was commonly used in Swedish hospital laboratories. These historical data on serology have not been used in the current study.

Figure 1

(A) Flowchart showing the study population (n=1600) after applying exclusion criteria. (B) Co-occurrence of antibodies against cyclic citrullinated peptide version 2 (anti-CCP2), any ACPA peptide reactivity, IgA RF, IgG RF and IgM RF among 1600 patients with rheumatoid arthritis (RA) from the EIRA cohort. No single autoantibody specificity was found in 294 patients. ACPA, antibodies against citrullinated protein; EIRA, Epidemiological Investigations in RA; IgG, immunoglobulin G; RF, rheumatoid factor; SRQ, Swedish Rheumatology Quality.

Clinical characteristics

Through linkage to SRQ, we collected information on CRP, ESR, SJC, TJC, DAS28, DAS28CRP, global disease activity registered by the patient (global VAS) and modified Health Assessment Questionnaire (HAQ) Score at RA diagnosis (baseline).13 14 Information on the number of fulfilled ACR 1987 classification criteria at the time of RA diagnosis was available for 885/1600 patients of whom 789 (89%) fulfilled four or more non-RF criteria.

Serological markers

In total, 16 individual ACPA reactivities were measured with a custom-made multiplex array (Thermo Fisher Scientific, ImmunoDiagnostics, Uppsala, Sweden).8 15 The citrullinated peptides represented deiminated sequences from filaggrin, fibrinogen, vimentin, a-enolase, collagen type II and hnRNP A3 as described in detail including peptide sequences.8 Fluorescence intensities were normalised and expressed as arbitrary units. For all patients and controls, fluorescence intensities for non-citrullinated control peptides were subtracted, except for CII359-369 (citC1) where the non-deiminated control antigen is in itself a conformation-dependent autoantigen where epitopes are destroyed by citrullination.16 Using the same sera, IgA, IgG and IgM RF isotypes were analysed with enzyme immune assay (EliA, Phadia AB, Uppsala, Sweden) according to the manufacturer’s instructions. Cut-offs for individual ACPA on the microarray and for RF isotypes were determined as the 98th percentile among 551 (individual ACPA reactivities) and 623 (RF isotypes) population control subjects without RA from the EIRA Study, respectively (IgA RF=18.01 IU/mL; IgM RF=9.96 IU/mL; IgG=34.36 µg/mL). Anti-CCP2 was analysed with ELISA (CCPlus, Eurodiagnostica, Malmö, Sweden). Results above 25 arbitrary units/mL were regarded as positive, in agreement with the manufacturer’s instructions. Among the 551 RA-free population controls, 9 were anti-CCP2 positive with this assay, yielding a specificity of 98.4%.


Descriptive clinical and serological data were tabulated. The association between the occurrence of individual serological markers, or combinations thereof, and the above-mentioned individual clinical phenotypic markers were assessed using linear regression adjusted for age (continuous), sex (binary), inclusion year (in quartiles, ordinal) and RA symptom duration (in days, continuous).3 5 17 18 Independent variables (serological markers) were treated as binary variables. ACPA positivity was alternatively determined as anti-CCP2 positivity or as occurrence of at least one individual ACPA peptide reactivity using the ACPA microarray. Dependent variables (clinical variables) were treated as continuous (SJC, TJC, ESR, CRP, patient global VAS, DAS28, DAS28CRP and HAQ).

The statistical calculations were performed with JMP V.16 (SAS Institute, Cary, North Carolina, USA). P values<0.05 were considered significant.


Characteristics of the study population

Table 1 describes the characteristics of the study population stratified based on presence of different autoantibody reactivities. The mean age at inclusion was 52 years, 1127 (70%) were women. The mean DAS28 at RA diagnosis was 5.3.

Table 1

Unadjusted mean values of RA signs and symptoms for patients positive and negative for individual autoantibodies or with different autoantibody combinations

Distribution of serological markers

Figure 1B and online supplemental table 1 display the distribution of positivity for each of the investigated serological markers and combinations thereof. In total, 1020 (64%) patients were positive for CCP2, 692 (43%) for IgA RF, 529 (33%) for IgG RF and 916 (57%) for IgM RF. Among these 1600 patients, 1243 (78%) individuals were positive for at least one of the ACPA subspecificities; the proportion of positives for each individual ACPA subspecificity varied from 10% to 57%. No further analyses concerning individual ACPA specificities were performed.

Supplemental material

Distribution of anti-CCP2 and any ACPA peptide reactivity by RF

There was a considerable overlap in positivity across the autoantibodies investigated (figure 1B). Online supplemental table 2 describes the overlap of anti-CCP2 (low, high and extreme), number of individual ACPA reactivities and RF overall and by RF isotype. Out of those positive for anti-CCP2, 88% were positive for at least one RF isotype, whereas 78% having at least one ACPA peptide were positive for at least one RF isotype.

The agreement between the occurrence of the compared RA autoantibodies varied from slight to substantial, with the highest kappa value for anti-CCP2/IgM RF (0.63) and the lowest for ACPA reactivity (0 vs ≥1 peptides)/IgG RF (0.20). Also, anti-CCP2 and any individual ACPA reactivity (0 vs ≥1 peptides) showed a high kappa value of 0.63; however, these measures of anticitrulline antibodies were otherwise never used in the same comparisons.

Within the eligible 364 days of symptom duration at diagnosis, 44% of the patients had >182 days duration. These patients had more individual ACPA reactivities than patients with shorter disease duration (median 7 vs 5 ACPA reactivities, p=0.0025). Generally, autoantibody-positive patients had somewhat longer symptom duration than autoantibody-negative patients (table 1). Consequently, all clinical evaluations were adjusted for symptom duration.

Exploratory regression analyses

In linear regression analysis where all markers individually were investigated for their association with the clinical symptoms at the time of RA diagnosis (model adjusted for age, sex, inclusion year and symptom duration), anti-CCP2, any ACPA peptide reactivity, IgA RF and IgM RF were all associated with lower SJC and TJC. IgG RF did not demonstrate any such association (table 2).

Table 2

Associations between anti-CCP2, ACPA positivity determined as any ACPA peptide reactivity, or individual RF isotypes, and signs and symptoms at the time of RA diagnosis among 1600 EIRA patients

All investigated autoantibodies were associated with higher ESR. In addition, all autoantibodies apart from IgG RF were also associated with higher CRP when adjusted for age, sex, inclusion year and symptom duration (table 2).

Totally unadjusted comparisons yielded similar results and are shown for the relation between involved joints and ACPA in online supplemental figure 1 A–D.

In multivariate analysis IgM RF associates with systemic inflammation and ACPA with low joint counts

In regression analyses with the three RF isotypes and any ACPA peptide reactivity investigated all together as independent variables (model adjusted for age, sex, inclusion year and symptom duration), individual clinical RA phenotypes associated with different autoantibodies.

IgM and IgA RF associated strongly with high ESR, whereas ACPA positivity associated with low SJC and TJC. ACPA peptide positivity was also associated with lower DAS28CRP (table 3). Similar associations were found when defining ACPA-positive group as having an anti-CCP2-positive test (online supplemental table 3).

Table 3

Regression with occurrence of IgA, IgG, IgM RF and ACPA positivity determined as any ACPA peptide reactivity evaluated together as independent variables and signs, symptoms or DAS28 evaluated individually as dependent variables

The strong association between ACPA, but not with RF, and low joint counts also remained when ACPA competed with individual RF isotypes, one at a time, in separate regression models (data not shown).

In order to corroborate these findings and to address the issue of agreement between the different RA autoantibodies, we performed a number of analysis where we stratified on IgA, IgM and IgG RF isotypes or ACPA positivity (determined as any peptide reactivity and as anti-CCP2 positivity, respectively) and evaluated how this was associated with levels of ESR and SJC (online supplemental table 4).

We found that RF positivity (of any isotype) associated with elevated ESR only when ACPAs (defined either as CCP2 positive or as having at least one individual ACPA) were present (online supplemental table 4, upper part). On the other hand, SJC was significantly lower in ACPA positives (defined by both ways) compared with ACPA-negative patients when RF was absent (all isotypes; online supplemental table 4, lower part). Stratification for RF isotypes significantly weakened the association between ACPA and ESR, and stratification for ACPA significantly weakened the effects of RF isotypes on the SJC (online supplemental table 4).

The association between RF and elevated ESR depends on the occurrence of ACPA

We further examined these associations in IgM RF-negative patients by adjusting for age, sex, inclusion year and symptom duration. Occurrence of ACPA associated not only with low SJC and TJC but also with low DAS28 and DAS28CRP; the association was strongest with joint counts irrespective of what ACPA measures were used (table 4). Corresponding data were obtained when we analysed patients negative for all RF isotypes (ie, not only those negative for IgM RF, online supplemental table 5).

Table 4

Association between ACPA positivity and signs and symptoms in the 684 IgM RF-negative patients among 1600 patients with RA from the EIRA cohort

These associations generally showed stronger significances among ever smokers than among never smokers (online supplemental table 6).

We next investigated the purely ACPA-negative group and defined ACPA-negative patients as having zero individual ACPA reactivities. In this group, we found no differences between patients with or without any of the three RF isotypes, for any phenotypic measures investigated. Notably, the strong association between IgM RF and elevated ESR found in previous analyses was completely absent in those who were negative for all ACPA reactivities (table 5). Comparable results were found when defining seronegativity as lacking anti-CCP2, although IgA and IgM RF associated with elevated ESR to some extent (online supplemental table 7).

Table 5

Association between RF positivity and signs and symptoms in the 357 totally ACPA peptide-negative patients among 1600 patients with RA from the EIRA cohort

When grouping the patients according to the number of ACPA reactivities, the association of RF with elevated ESR increased with the number of individual ACPA reactivities present and was most prominent for IgM RF, less so for IgA RF and for IgG RF was only seen for the group with the highest number of ACPA reactivities (figure 2A–C). This evaluation was complicated by the fact that groups with the highest number of ACPA reactivities also had somewhat higher levels of RF isotypes (online supplemental table 8).

Figure 2

(A–C) Associations between ESR on the one hand and IgA RF, IgG RF and IgM RF stratified for the number of individual ACPA reactivities on the other hand. Graphs are based on least square mean levels from regression analyses, with all models adjusted for age, sex, inclusion year and symptom duration. Lines above the bars indicate SE of the mean. ACPA, antibodies against citrullinated protein; ESR, erythrocyte sedimentation rate; IgA, immunoglobulin A; RF, rheumatoid factor.

The associations between IgM RF and high ESR and between ACPA positivity and low SJC are not an effect of the ACR classification criteria

When evaluating the 789 patients that fulfilled four or more non-RF ACR 1987 criteria, the associations between IgM RF and high ESR/high CRP remained (table 6, left part). Corresponding results were seen when comparing ACPA peptide-positive and ACPA peptide-negative patients in this subgroup where associations with both elevated ESR and CRP, as well as with low joint counts, remained (table 6, right part).

Table 6

Comparison between IgM RF status (left part) and ACPA status determined as any ACPA peptide reactivity (right part) in all patients versus patients fulfilling four or more non-RF ACR criteria


In our study, presence of ACPA associated with lower number of swollen and tender joints in the RF-negative group, whereas presence of IgM RF associated with increased inflammatory markers, especially ESR, dependent on the co-occurrence of ACPA. In other words, in early RA, RF and ACPA associate with different individual DAS28 components and in opposite directions.

The association between ACPA and lower joint counts was shown in regression analyses including both ACPA and RFs (table 3), as well as in analyses stratified by the occurrence of RF isotypes (table 4 and online supplemental tables 4 and 5). The association between RF and elevated levels of inflammatory markers, especially ESR, is in agreement with previous studies.6 19

It is tempting to speculate that the ACPA-associated early phenotype with lower joint counts implies that the early prediagnostic pathology of ACPA-positive arthritis might start locally in a few or even in a single joint, in a way different from that of ACPA-negative arthritis. As shown previously in other studies, the ACPA response is initially restricted with more individual ACPA developing over time and especially close to disease onset.20 In agreement with this, we found more ACPA reactivities in patients with longer symptom duration within the eligible 364 days and adjusted all analyses for symptom duration.

Previous studies by Barra et al and Nordberg et al have reported lower DAS28 and fewer swollen and tender joints in newly diagnosed seropositive patients.4 5 Both these studies defined seropositivity as having either RF or ACPA, although Barra et al performed subanalysis without finding any phenotypic difference between patients with different RF/ACPA combinations. In that study, IgM RF analyses performed locally in different centres had been pooled and ACPAs were not consistently analysed, that is, different assays performed in different laboratories were used.4 In our study, all autoantibody analyses had been performed centrally with one technique per analyte and all analytes were given cut-offs with similar specificities in relation to the same control group, allowing us to study the effect of ACPA in RF-negative subjects and vice versa. By employing data from an ACPA multiplex array, we could carefully evaluate the effect of gradual addition of ACPA specificities on the RF-associated inflammatory response. Notably, the ACPA multiplex array allowed us to define a clean ACPA-negative subset devoid of the few ACPA-positive subjects found among anti-CCP2-negative patients.8 This is in agreement with a recent study showing residual ACPA reactivity among anti-CCP2-negative and RF-positive individuals with increased levels of proinflammatory cytokines.21

Nordberg et al suggested the lower degree of inflammation among seropositive patients to be due to the 2010 classification criteria requiring more inflamed joints for diagnosis of seronegative RA. In our study, all patients were classified according to the 1987 criteria only including RF. When investigating patients fulfilling four or more non-RF criteria, that is, fulfilling the ACR criteria even without autoantibody serology, we showed that the phenotypic differences between IgM RF or ACPA-positive and ACPA-negative patients, respectively, remained (table 6). We therefore believe that our results describe biologically determined phenotypes rather than an effect of classification criteria. That said, we have no formal means to rule out the possibility that the primary care physicians’ awareness of IgM RF status might have influenced the referral to rheumatology and subsequent enrolment in EIRA. Similarly, there is a possibility that knowledge of IgM RF status might have influenced therapy given by primary care physicians at the time of referral to a rheumatologist and before RA diagnosis. Currently and during the study period, RA diagnoses in Sweden are and were more or less exclusively assigned by rheumatologists, based on referrals from primary care physicians to rheumatology departments. The general instruction to primary care is, and has been, to avoid the use of oral or other corticosteroids in patients referred to rheumatology for rapid diagnostic work-up. Similarly, in Sweden, disease-modifying antirheumatic drugs (DMARDs) are exclusively started and managed by rheumatologists. Therefore, we do not think that prediagnostic therapy might have differed considerably between seropositive and seronegative patients in the current study.

In our study, IgM RF positivity did not associate with ESR among patients without any ACPA reactivity, but gradually appeared with the more ACPA specificities expressed by the patients. We showed this in the following different ways: by analysing ESR in ACPA-negative subjects (table 5 and online supplemental table 7), evaluating the effect of RFs in different strata of individual ACPA (online supplemental table 4 and figure 2A–C) and in multiple regression (table 3). Collectively, these analyses argue that IgM RF is clearly associated with elevated ESR at the time of RA diagnosis, but only in the context of ACPA. These findings might be viewed in relation to laboratory in vitro data on ACPA function. The laboratories in Toulouse and Stanford developed techniques to construct surface-bound immune complexes (IC) containing citrullinated fibrinogen and ACPA and showed that such IC stimulated production of tumour necrosis factor from macrophages.22 23 Later studies showed that addition of RF enhanced the cellular responses to surface-bound ACPA IC.24 25 When the initial ACPAs were excluded from these experimental systems, RF induced only marginal levels of tumour necrosis factor (TNF), arguing that this mechanism for RF-associated inflammation is ACPA dependent, which is in agreement with our clinical findings in this paper.24 25 Previous studies have shown that double RF/ACPA positivity is associated with higher levels of cytokines, stronger prediction and earlier appearance of clinical RA in at-risk individuals and higher probability of appearance of interstitial lung disease in patients with RA.21 26–28 Our clinical data showing an ACPA-dependent proinflammatory effect of RF are therefore compatible with the hypothesis that immobilised ACPA-containing IC induce enhanced levels of TNF and interleukin 6 (IL-6) after addition of RF, but that RF in itself has no proinflammatory effect in that model in vivo (figure 3).

Figure 3

Hypothetical figure showing the ACPA dependency of RF-enhanced production of TNF after formation of immobilised IC between ACPA and RF (left) versus after the formation of solubilised IC with low TNF-inducing capacity when ACPAs are not present (right). The soluble IC might or might not contain ACPA. Created with ACPA, antibodies against citrullinated protein; IC, immune complexes; IgM, immunoglobulin M; RF, rheumatoid factor; TNF, tumour necrosis factor.

We find that the effect of IgM RF is larger on ESR than on CRP. In a previous study, we noted that there was a larger difference in ESR than in CRP between anti-CCP2-positive and anti-CCP2-negative EIRA patients, see figure 2a and b in Manivel et al.29 Therefore, we think and hypothesise that this difference is an inherent property of the EIRA patient cohort rather than reflecting a difference between how the laboratory markers are associated to IgM RF.

Three previous studies have presented data arguing that some ACPA specificities might have anti-inflammatory properties as shown from the demonstration that certain monoclonal RA-derived ACPAs may counteract development of experimental arthritis induced in different ways.30–32 These data may be of relevance for interpretation of our findings of low joint swollen count being associated with ACPA positivity in early RA. A precaution is that the anti-inflammatory effects of some ACPA have so far been described only in certain mouse models and also that proinflammatory monoclonal ACPAs were identified in one of these studies32 indicating that there may be a large variability in functions between different IgG ACPA clones with different fine specificities. If this would represent a true mechanism behind our findings, it may be that a change towards more proinflammatory ACPA specificities occurs later in the disease course, as presence of ACPA is associated with more swollen joints in long-standing RA.3 Future studies on the longitudinal changes of ACPA repertoires may approach this issue.

The co-occurrence and the strong biological interaction between the ACPAs and RFs may be identified as a weakness of the study, since it makes it challenging to disentangle their individual contributions to RA pathology. We have tried to address this problem by using a combination of both adjusted and stratified approaches. Another limitation is the lack of data on erosions in EIRA. By contrast, important strengths of our study are the large homogenous patient group with limited time between sampling for autoantibody measurements and recording of the clinical phenotype. Another strength is that all autoantibodies were measured centrally with a single assay per analyte and that all analytes were given similar diagnostic specificity.

In conclusion, we have shown that in early RA, ACPAs associate with lower counts of affected joints and RF associates with elevated measures of systemic inflammation in an ACPA-dependent manner. Thus, future studies and predictive algorithms in early RA might benefit from individual assessment of the impact of different autoantibodies as well as separate evaluation of individual components of DAS28.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and was approved by the ethics board in Stockholm (EPM registration number 2021-00125). Participants gave informed consent to participate in the study before taking part.


This study was performed within the Nordic collaborative project NORA.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Handling editor Josef S Smolen

  • Twitter @MartinCornillet

  • Contributors EP and JR conceived the study, had full access to all data, made statistical calculations and drafted the manuscript together with JA. LA, LK, SS and HW provided epidemiological and clinical data and expertise. MH and LM-A performed autoantibody measurements. MC, P-JJ, RH, KS and GS provided peptides for the analyses of specific antibodies against citrullinated proteins. VAM performed statistical analyses. All authors read, commented on and approved the final manuscript. JR is the guarantor of this study.

  • Funding This study was supported by funding (to LK, LA, RH, JA, JR) from Vinnova, Innovationsfonden and the Research Council of Norway under the frame of NordForsk (grant agreement number 90825, Project NORA), by the Swedish Research Council, the Federal Ministry of Education and Research in Germany and the Research Council of Norway, under the frame of ERA PerMed (Project ScandRA), the Swedish Foundation for Strategic Research, Sweden's innovation agency, the Knut and Alice Wallenberg Foundation, the Swedish Research Council for Health, Working Life and Welfare, the insurance company AFA, the Swedish Rheumatism Association, King Gustaf V:s 80-year Foundation and the Stockholm County Council.

  • Competing interests EP and LM-A are employees at Thermo Fisher Scientific. JA has had or have research agreements with AbbVie, BMS, Eli Lilly, Galapagos, MSD, Pfizer, Roche, Samsung Bioepis and Sanofi, mainly for the national safety monitoring of rheumatology immunomodulators in Sweden (ARTIS). JR has been a member of the Scientific Advisory Board for Thermo Fisher Scientific and for Inova/Werfen and has received consulting fees, speaking fees and/ or honoraria by Thermo Fisher Scientific. RH has received consulting fees from Regor, Lipum AB and Cyxone AB and is the founder of Vacara AB. LK has been a co-ordinator of several IMI-funded projects which included collaborations with Janssen, Pfizer, Sanofi, UCB, GSK and Eli-Lilly. This work has been presented as a poster in a preliminary form at EULAR (June 2019) and to the European Workshop for Rheumatology Research (March 2023).

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.